How to Make a Chart in Tableau with AI

Creating charts in Tableau used to be all about mastering the drag-and-drop interface, knowing which shelf to place your pills on, and remembering the difference between discrete and continuous fields. Now, you can build compelling visualizations simply by describing what you want to see. This guide will walk you through exactly how to make charts in Tableau using its new AI features, step-by-step, no expert knowledge required.

Graphed

Build AI Agents for Marketing

Build virtual employees that run your go to market. Connect your data sources, deploy autonomous agents, and grow your company.

Build Your First Agent

What is Tableau AI (and Einstein Copilot)?

Tableau AI isn't just one feature, it's a suite of artificial intelligence capabilities integrated directly into the Tableau platform. The most visible part of this is Einstein Copilot for Tableau, a conversational AI assistant that lives right inside your workspace. Think of it as having a Tableau expert sitting next to you, ready to build charts, perform calculations, and answer questions about your data based on your instructions in plain English.

The goal is to lower the barrier to entry for data analysis. Instead of spending hours learning the technical nuances of the software, you can focus on the questions you want to answer. You describe the chart you need, and the AI handles the technical steps of building it, translating your words into a finished visualization.

Before You Begin: Getting Your Data Ready for AI

While Tableau's AI is powerful, it works best with data that is clean, organized, and easy for both humans and machines to understand. The saying "garbage in, garbage out" still applies. Before you start prompting the AI, spending a few minutes on your data source will save you a lot of time later.

Connect Your Data Source

This first step hasn't changed. You still need to connect Tableau to your data, whether it's an Excel spreadsheet, a SQL database, or a cloud service. Open your Tableau workbook and connect to the data you want to analyze. Once it's loaded, you'll see your data fields in the "Data" pane on the left, just like you normally would.

AI Agents for Marketing

Crash Course

Graphed

Free PDF · the crash course

AI Agents for Marketing Crash Course

Learn how to deploy AI marketing agents across your go-to-market — the best tools, prompts, and workflows to turn your data into autonomous execution without writing code.

Download the PDF

Check for Clear Column Names

This is probably the single most important step for getting good results from the AI. Einstein Copilot understands context, but it relies on your column headers to know what each field represents. Your column names should be descriptive and clear.

• Good Example: Customer Name, Order Date, Sales Revenue, Shipping Cost, Product Category
• Bad Example: cust_n, orddt, value1, ship_#, prod_cat_id

If your column names are cryptic, rename them in the Data Source tab before you begin. A clear name like "Sales Revenue" makes it easy to ask, "Show me my sales revenue," without confusing the AI.

Understand Your Data Structure

Briefly scan your data to understand what it contains. What are the key metrics (like sales, profit, quantity)? What are the dimensions or categories you can use to slice and dice those metrics (like region, product category, customer segment)? Having a basic mental map of your data helps you ask better questions.

A Step-by-Step Guide to Creating Your First AI-Powered Chart

Let's walk through an example using a common dataset: sales information for a fictional store. Assume our data has fields like Order Date, Region, Product Category, Sales, and Profit.

Step 1: Open the Einstein Copilot Pane

When you open a worksheet in a recent version of Tableau, you'll see an Einstein icon in the toolbar. Clicking this will open the conversational AI pane. This is your command center for building with AI. It includes a text box where you'll type your prompts.

Step 2: Write Your First Prompt

Start with a simple and direct request. For your first chart, don't try to get too complex. The goal is to get a basic visualization built that you can refine later. Let's ask the AI to create a simple bar chart.

In the Einstein Copilot text box, type:

Show me total sales by product category as a bar chart

And hit enter. That's it.

Graphed

Build AI Agents for Marketing

Build virtual employees that run your go to market. Connect your data sources, deploy autonomous agents, and grow your company.

Build Your First Agent

Step 3: Analyze the Results

Tableau AI will get to work instantly. It parses your request, identifies the fields (Sales and Product Category), and understands the aggregation (total) and the desired chart type (bar chart). In seconds, you’ll see the bar chart appear on your canvas. It will have automatically placed Product Category on the Columns shelf and SUM(Sales) on the Rows shelf. The AI handled all the dragging and dropping for you.

Step 4: Refine Your Chart with Follow-up Questions

This is where the magic really happens. A first draft is nice, but data analysis is an iterative process. You can continue the conversation with Einstein Copilot to modify and improve your chart without a single click.

Let's refine our bar chart with a few follow-up prompts:

• To sort the data: Type, "Sort this in descending order." The AI will instantly apply a sort to the Product Category based on sales, putting the highest-performing category first.
• To add color and more context: Try something like, "Now color the bars by profit." The Copilot will figure out that you want to add the SUM(Profit) field to the Color mark, allowing you to see which best-selling categories are also the most profitable (and which aren't).
• To add a filter: If you only want to see data for one area, you can say, "Filter this to only show the West region." A filter for Region will be created and applied automatically.
• To change the chart type: Maybe a bar chart isn't the best view. Just ask for a new one. "Change this to a treemap." The view will update instantly, preserving your existing fields and filters.

With just a few simple sentences, you've created, sorted, colored, filtered, and completely changed the structure of your visualization. This conversational back-and-forth makes exploration incredibly fast and intuitive.

Tips and Tricks for Writing Better AI Prompts

The better your prompts, the better your results. While you can be conversational, being clear and specific goes a long way. Here are some tips to get the most out of Tableau AI.

Be Specific and Clear

Ambiguity is the enemy of good AI output. A vague prompt will often lead to a result, but it might not be the one you want.

• Instead of: "Show me sales."
• Try: "Create a line chart tracking monthly sales trends over the last two years."

The second prompt specifies the chart type (line chart), the metric (sales), the dimension for the x-axis (monthly time trend), and the timeframe (last two years). This leaves very little room for misinterpretation.

Use Natural Questions

You don't need to learn a special query language. Frame your requests as questions you would ask a human analyst. This approach often leads to insightful visualizations because it's rooted in a real business question.

• "Which five states generate the most profit?"
• "What is the relationship between shipping cost and profit?" (This might generate a scatter plot)
• "Show me the year-over-year sales growth as a percentage."

AI Agents for Marketing

Crash Course

Graphed

Free PDF · the crash course

AI Agents for Marketing Crash Course

Learn how to deploy AI marketing agents across your go-to-market — the best tools, prompts, and workflows to turn your data into autonomous execution without writing code.

Download the PDF

Explicitly Ask for Chart Types

While the AI can often infer a good chart type, you have full control. If you know what you want, just ask for it directly. Common chart types you can request include:

• bar chart
• line chart
• pie chart
• scatter plot
• map
• treemap
• heatmap

For example, you could ask: "Display profit by state on a map."

Build Complex Charts Incrementally

Instead of trying to build a highly complex chart with one very long sentence, build it piece by piece. Start with a foundational chart (e.g., "sales by region") and then layer on details with follow-up prompts:

1. "Show me sales by region in a bar chart."
2. "Now, break down each bar by product category."
3. "Color this by shipping mode."
4. "Add average profit as a label on each category segment."

This iterative process allows you to check the AI's work at each step and makes it easier to troubleshoot if one of your prompts doesn't produce the expected result.

Handling Common Problems

Sometimes the AI might get stuck or misunderstand your request. Here’s how to deal with common issues.

• The AI Misunderstands: If the chart isn't what you expected, simply rephrase your prompt. Try being more specific or using slightly different words. For instance, if "Display sales figures" didn't work, try "Create a chart showing the sum of sales."
• Data Doesn't Seem Right: Tableau AI doesn't change your underlying data, it only visualizes it. If the numbers look wrong, the issue is almost always in the source data itself. Go back to the Data Source tab to check for errors, duplicates, or missing values.
• AI Can't Find a Field: Remember to use the actual names of the fields in your data pane. If your data has a field called Cust_Segment, asking for "customer groups" won't work. Being precise with field names solves this problem immediately.

Final Thoughts

By integrating conversational AI like Einstein Copilot, Tableau has made the process of creating charts significantly faster and more accessible. Now, anyone can turn data into meaningful visualizations by asking questions in plain English, allowing you to move directly from a business question to a visual insight without getting bogged down in the technical details of the tool.

While Tableau's AI is incredibly powerful for users already within that ecosystem, the real challenge for many teams is getting all their data in one place to begin with. To solve this, we built Graphed target="_blank" rel="noopener"). It lets you connect all your scattered marketing and sales data - from platforms like Google Analytics, Shopify, Facebook Ads, and Salesforce - in seconds. From there, you can build entire dashboards and get instant answers using simple, natural language, giving you a full, cross-platform view of your business without any complex setup or manual reporting.